Artificial intelligence (AI) permeates all fields of life, which resulted in new challenges in requirements engineering for artificial intelligence (RE4AI), e.g., the difficulty in specifying and validating requirements for AI or considering new quality requirements due to emerging ethical implications. It is currently unclear if existing RE methods are sufficient or if new ones are needed to address these challenges. Therefore, our goal is to provide a comprehensive overview of RE4AI to researchers and practitioners. What has been achieved so far, i.e., what practices are available, and what research gaps and challenges still need to be addressed? To achieve this, we conducted a systematic mapping study combining query string search and extensive snowballing. The extracted data was aggregated, and results were synthesized using thematic analysis. Our selection process led to the inclusion of 126 primary studies. Existing RE4AI research focuses mainly on requirements analysis and elicitation, with most practices applied in these areas. Furthermore, we identified requirements specification, explainability, and the gap between machine learning engineers and end-users as the most prevalent challenges, along with a few others. Additionally, we proposed seven potential research directions to address these challenges. Practitioners can use our results to identify and select suitable RE methods for working on their AI-based systems, while researchers can build on the identified gaps and research directions to push the field forward.

Large Language Models (LLMs) have revolutionized numerous applications, making them an integral part of our digital ecosystem. However, their reliability becomes critical, especially when these models are exposed to misinformation. We primarily analyze the susceptibility of state-of-the-art LLMs to factual inaccuracies when they encounter false information in a QnA scenario, an issue that can lead to a phenomenon we refer to as *knowledge drift*, which significantly undermines the trustworthiness of these models. We evaluate the factuality and the uncertainty of the models' responses relying on Entropy, Perplexity, and Token Probability metrics. Our experiments reveal that an LLM's uncertainty can increase up to 56.6% when the question is answered incorrectly due to the exposure to false information. At the same time, repeated exposure to the same false information can decrease the models uncertainty again (-52.8% w.r.t. the answers on the untainted prompts), potentially manipulating the underlying model's beliefs and introducing a drift from its original knowledge. These findings provide insights into LLMs' robustness and vulnerability to adversarial inputs, paving the way for developing more reliable LLM applications across various domains. The code is available at https://github.com/afastowski/knowledge_drift.

Knowledge conflict arises from discrepancies between information in the context of a large language model (LLM) and the knowledge stored in its parameters. This can hurt performance when using standard decoding techniques, which tend to ignore the context. Existing test-time contrastive methods seek to address this by comparing the LLM's output distribution with and without the context and adjust the model according to the contrast between them. However, we find that these methods frequently misjudge the degree of conflict and struggle to handle instances that vary in their amount of conflict, with static methods over-adjusting when conflict is absent. We propose a fine-grained, instance-level approach called AdaCAD, which dynamically infers the weight of adjustment based on the degree of conflict, as measured by the Jensen-Shannon divergence between distributions representing contextual and parametric knowledge. Our experiments across four models on six diverse question-answering (QA) datasets and three summarization tasks demonstrate that our training-free adaptive method consistently outperforms other decoding methods on QA, with average accuracy gains of 14.21% (absolute) over a static contrastive baseline, and improves the factuality of summaries by 5.59 (AlignScore). Furthermore, our analysis shows that while decoding with contrastive baselines hurts performance when conflict is absent, AdaCAD mitigates these losses, making it more applicable to real-world datasets in which some examples have conflict and others do not.

This article presents a method for verifying RDF triples using LLMs, with an emphasis on providing traceable arguments. Because the LLMs cannot currently reliably identify the origin of the information used to construct the response to the user query, our approach is to avoid using internal LLM factual knowledge altogether. Instead, verified RDF statements are compared to chunks of external documents retrieved through a web search or Wikipedia. To assess the possible application of this workflow on biosciences content, we evaluated 1,719 positive statements from the BioRED dataset and the same number of newly generated negative statements. The resulting precision is 88%, and recall is 44%. This indicates that the method requires human oversight. We demonstrate the method on Wikidata, where a SPARQL query is used to automatically retrieve statements needing verification. Overall, the results suggest that LLMs could be used for large-scale verification of statements in KGs, a task previously unfeasible due to human annotation costs.

Russian Internet Trolls use fake personas to spread disinformation through multiple social media streams. Given the increased frequency of this threat across social media platforms, understanding those operations is paramount in combating their influence. Using Twitter content identified as part of the Russian influence network, we created a predictive model to map the network operations. We classify accounts type based on their authenticity function for a sub-sample of accounts by introducing logical categories and training a predictive model to identify similar behavior patterns across the network. Our model attains 88% prediction accuracy for the test set. Validation is done by comparing the similarities with the 3 million Russian troll tweets dataset. The result indicates a 90.7% similarity between the two datasets. Furthermore, we compare our model predictions on a Russian tweets dataset, and the results state that there is 90.5% correspondence between the predictions and the actual categories. The prediction and validation results suggest that our predictive model can assist with mapping the actors in such networks.

Large Vision-Language Models (LVLMs), trained on multimodal big datasets, have significantly advanced AI by excelling in vision-language tasks. However, these models remain vulnerable to adversarial attacks, particularly jailbreak attacks, which bypass safety protocols and cause the model to generate misleading or harmful responses. This vulnerability stems from both the inherent susceptibilities of LLMs and the expanded attack surface introduced by the visual modality. We propose Sim-CLIP+, a novel defense mechanism that adversarially fine-tunes the CLIP vision encoder by leveraging a Siamese architecture. This approach maximizes cosine similarity between perturbed and clean samples, facilitating resilience against adversarial manipulations. Sim-CLIP+ offers a plug-and-play solution, allowing seamless integration into existing LVLM architectures as a robust vision encoder. Unlike previous defenses, our method requires no structural modifications to the LVLM and incurs minimal computational overhead. Sim-CLIP+ demonstrates effectiveness against both gradient-based adversarial attacks and various jailbreak techniques. We evaluate Sim-CLIP+ against three distinct jailbreak attack strategies and perform clean evaluations using standard downstream datasets, including COCO for image captioning and OKVQA for visual question answering. Extensive experiments demonstrate that Sim-CLIP+ maintains high clean accuracy while substantially improving robustness against both gradient-based adversarial attacks and jailbreak techniques. Our code and robust vision encoders are available at https://github.com/speedlab-git/Robust-Encoder-against-Jailbreak-attack.git.

In the past decade, social media platforms have been used for information dissemination and consumption. While a major portion of the content is posted to promote citizen journalism and public awareness, some content is posted to mislead users. Among different content types such as text, images, and videos, memes (text overlaid on images) are particularly prevalent and can serve as powerful vehicles for propaganda, hate, and humor. In the current literature, there have been efforts to individually detect such content in memes. However, the study of their intersection is very limited. In this study, we explore the intersection between propaganda and hate in memes using a multi-agent LLM-based approach. We extend the propagandistic meme dataset with coarse and fine-grained hate labels. Our finding suggests that there is an association between propaganda and hate in memes. We provide detailed experimental results that can serve as a baseline for future studies. We will make the experimental resources publicly available to the community.

Recent state-of-the-art authorship attribution methods learn authorship representations of texts in a latent, non-interpretable space, hindering their usability in real-world applications. Our work proposes a novel approach to interpreting these learned embeddings by identifying representative points in the latent space and utilizing LLMs to generate informative natural language descriptions of the writing style of each point. We evaluate the alignment of our interpretable space with the latent one and find that it achieves the best prediction agreement compared to other baselines. Additionally, we conduct a human evaluation to assess the quality of these style descriptions, validating their utility as explanations for the latent space. Finally, we investigate whether human performance on the challenging AA task improves when aided by our system's explanations, finding an average improvement of around +20% in accuracy.

Following the Russian Federation's full-scale invasion of Ukraine in February 2022, a multitude of information narratives emerged within both pro-Russian and pro-Ukrainian communities online. As the conflict progresses, so too do the information narratives, constantly adapting and influencing local and global community perceptions and attitudes. This dynamic nature of the evolving information environment (IE) underscores a critical need to fully discern how narratives evolve and affect online communities. Existing research, however, often fails to capture information narrative evolution, overlooking both the fluid nature of narratives and the internal mechanisms that drive their evolution. Recognizing this, we introduce a novel approach designed to both model narrative evolution and uncover the underlying mechanisms driving them. In this work we perform a comparative discourse analysis across communities on Telegram covering the initial three months following the invasion. First, we uncover substantial disparities in narratives and perceptions between pro-Russian and pro-Ukrainian communities. Then, we probe deeper into prevalent narratives of each group, identifying key themes and examining the underlying mechanisms fueling their evolution. Finally, we explore influences and factors that may shape the development and spread of narratives.

Background: Health 3.0 allows decision making to be based on longitudinal data from multiple institutions, from across the patient's healthcare journey. In such a distributed setting, blockchain smart contracts can act as neutral intermediaries to implement trustworthy decision making. Objective: In a distributed setting, transmitted data will be structured using standards (such as HL7 FHIR) for semantic interoperability. In turn, the smart contract will require interoperability with this standard, implement a complex communication setup (e.g., using oracles), and be developed using blockchain languages (e.g., Solidity). We propose the encoding of smart contract logic using a high-level semantic Knowledge Graph, using concepts from the domain standard. We then deploy this semantic KG on blockchain. Methods: Off-chain, a code generation pipeline compiles the KG into a concrete smart contract, which is then deployed on-chain. Our pipeline targets an intermediary bridge representation, which can be transpiled into a specific blockchain language. Our choice avoids on-chain rule engines, with unpredictable and likely higher computational cost; it is thus in line with the economic rules of blockchain. Results: We applied our code generation approach to generate smart contracts for 3 health insurance cases from Medicare. We discuss the suitability of our approach - the need for a neutral intermediary - for a number of healthcare use cases. Our evaluation finds that the generated contracts perform well in terms of correctness and execution cost ("gas") on blockchain. Conclusions: We showed that it is feasible to automatically generate smart contract code based on a semantic KG, in a way that respects the economic rules of blockchain. Future work includes studying the use of Large Language Models (LLM) in our approach, and evaluations on other blockchains.